Fast and reliable function calling is critical to building highly capable AI copilots and assistants. In this deep dive, we will explore the lessons learned from building Gorilla, the leading open source large language model that integrates with a wide range of APIs, and the Berkeley Function-Calling Leaderboard (BFCL), the first comprehensive and executable function calling evaluation for LLMs.

Key takeaways include:

• The nuances of integrating LLMs with application APIs.
• Best practices for fine-tuning LLMs for function calling capabilities.
• The future of open source large language models and agents.

We’ll be joined by ​​Shishir Patil, a coauthor of Gorilla, Ph.D. student from UC Berkeley, and member of the Sky Computing Lab and Berkeley AI Research. With experience from Google Brain, Amazon Science Core ML, and Microsoft Research India, Shishir brings a wealth of knowledge and practical insights to this cutting-edge topic.